This post has been inspired, in some small amount, by IBM’s suggestion that businesses are unable to analyze over 90% of their data.
Jeff Jonas, IBM chief scientist for the entity and analytics group, has suggested that "as computers are getting faster, organizations are getting dumber - they are now lucky to fully understand seven percent of their data and this is steadily getting worse."
The data analytics equivalent to the age old question, “What comes first, the chicken or the egg?” is “What comes first, the analytic solution or the business requirements.
For a long time the conventional wisdom in telecoms circles has been that we need to know the questions the communication services providers (CSPs) want to answer before we turn them loose to analyze the data. That’s all changed. If we look back over just the past year and assess the plethora of attention being directed at big data and business analytics, we’ll find that a large part of the effort is about getting the data in front of the users faster and reducing the requirement gathering, data modelling, report building, and other steps that normally would take place before the business user ever saw the data she needed. For CSPs, in particular, discovering the insight hidden in plain sight, as it were, could make all the difference in minimising revenue leakage from a notoriously fickle customer base.
It raises the question of what CSPs should be looking for in their business data. Most people would reflexively say they should get answers.
I would propose, however, that CSPs should view data analytics more as a way to find out what the questions should be; letting the data guide business decisions rather than simply answering predetermined and recurring questions. For example, rather than assuming revenue leakage necessarily stems from traditional sources - such as billing inaccuracies or rising commission costs - and seeking to find data that matches the assumption, CSPs should let the data provide the lead for what the problem might be, before defining the solution. This intuition and discovery-driven approach to data analytics can be tightly intertwined with a specific business process, often leading to near real-time insights and insights that are easier to understand and act on to modify performance.
Even more importantly, however, with a discovery-driven approach, CSPs will find new areas where data analytics can improve performance, including areas of the business, such as those dealing with fraud or customer management, where new questions arise every day that go beyond pre-modeled and predetermined business logic.
So a discovery-driven approach, where analytic solutions are created before the full business requirements are known, can guide you to what you should be asking, give you the answers to these previously unknown questions and lead to more actionable insights that those provided by traditional reporting and data management processes.